ms. elizabeth m. murphy secretary washington, dc …...1 november 11, 2011 ms. elizabeth m. murphy...

November 11, 2011

Ms. Elizabeth M. Murphy Secretary U.S. Securities and Exchange Commission 100 F Street NE Washington, DC 20549-1090

RE: Center On Executive Compensation Comments on the Use of Statistical Sampling to Implement Section 953(b) of the Dodd-Frank Wall Street Reform and Consumer Protection Act

Dear Ms. Murphy:

The Center On Executive Compensation (“Center”) is pleased to submit comments to the U.S. Securities and Exchange Commission (“SEC” or “Commission”) regarding the implementation of Section 953(b) of the Dodd-Frank Wall Street Reform and Consumer Protection Act. That section requires all companies to calculate the median compensation of all employees other than the CEO as calculated under Item 402 of Regulation S-K under the Securities Exchange Act of 19341 and disclose that number and the ratio of the median employee’s pay to total compensation of the CEO in filings made with the Commission.

We understand from public comments recently made by SEC officials that the staff is giving consideration to a suggestion to implement the pay ratio requirement by allowing companies to determine the median compensation of all employees by using statistical sampling of the cash compensation of all employees in an attempt to reduce the compliance burden. While the Center is fully supportive of a flexible interpretation of section 953(b)—in fact it believes it essential for the reasons we have indicated in our previous submissions—sampling would not address the underlying issues that make the collection and analysis of global compensation so difficult and costly. Although sampling may save time in certain respects (e.g., by limiting the requirement that the compensation of all employees be calculated), it would introduce additional complexity by requiring the development of a methodology to determine the appropriate stratification of the sample population, develop and assess the appropriate confidence intervals to enhance the reliability of the data collected and ensure that comparable forms of compensation are included across the varying pay practices that are common in different regions of the world.

To determine the impact of sampling, we undertook a survey of Center Subscribers and HR Policy Association members. The results of the survey are presented below and indicate that identifying the information required to conduct sampling would be extremely difficult.

For these reasons, the Center urges the staff to undertake a thorough study of the suggestion not to implement the pay ratio through sampling and to consider various alternative approaches

1 Section 953(b) requires the ratio to be calculated in accordance with under Item 402 of Regulation S-K under the Securities Exchange Act of 1934 as in effect on July 21, 2010.

mblanchard

New Header

Ms. Elizabeth M. Murphy November 11, 2011

to implementation that truly reduce cost, complexity and difficulty and facilitate the ability of companies to comply with any rule ultimately adopted by the Commission.

As you know, the Center On Executive Compensation is a research and advocacy organization that seeks to provide a principles-based approach to executive compensation policy from the perspective of the senior human resource officers of leading companies. The Center is a division of HR Policy Association, which represents the chief human resource officers of over 330 large companies, and the Center’s more than 90 subscribing companies are HR Policy members that represent a broad cross-section of industries.

Statistical Sampling Would Be Extremely Difficult for Most Global Companies to Implement

In crafting regulations to implement section 953(b) of the Dodd-Frank Act, the Commission is required to consider the costs and benefits of its proposed approach and overall impact of its proposed regulation on issuers and competition as well as the benefit of the information to investors.2 Although the Center fully supports a flexible interpretation of section 953(b), the use of statistical sampling is not the solution. Statistical sampling would not ease the core burdens associated with calculating the pay ratio for most global companies that stem from aggregating global employee data.

Most companies, especially larger multinational companies, do not maintain a centralized list of employees that is linked to their compensation information. Instead, information about each individual employee, including payroll data, is typically maintained in each country, business unit or individual location where the company does business. Thus, to develop a representative sample, whether by targeting a certain percentage of all employees, seeking to achieve a certain confidence level, or using stratified sampling by country, region, location or division, companies would still need to develop a list that identified all employees and once the appropriate sample is selected, identify and calculate all the elements of compensation to be used for sampling purposes. This would be inordinately complex and time consuming, and as discussed below, many issuers are concerned that the calculations would not result in an accurate computation of the median estimates of all employees.

Survey Results Demonstrate That Sampling Would Not Ease Administrative or Financial Burdens of Compliance Because of the Wide Variability in Pay Practices and Recordkeeping

To demonstrate the difficulty large companies would have in complying with a pay ratio calculation that involved statistical sampling, the Center recently sponsored of survey of its Subscribers and other members of its parent organization, HR Policy Association. A total of 95 companies responded and many explained their multiple choice answers in detail. The full survey is attached, and the key findings below demonstrate why aggregation of the data, even on a sampling basis, is so complicated.

2 The Commission is required to consider the effect of its rules on “efficiency, competition, and capital formation,” 15 U.S.C. §§ 78c(f), 78w(a)(2), 80a-2(c).


Demographic Information

The survey asked questions about the number of employees and the number of countries and locations in which companies operate, as well as how many divisions and subsidiaries they have. The purpose of these questions was to provide a greater understanding of the scope of company operations and the amount of work that would be required to identify an appropriate sample across business units or locations.

Number of Employees. Over three-quarters of the respondents, 78.7%, have over 10,000 employees globally and over a third, 37.2% have over 50,000 employees globally.

Number of Countries. Three-quarters, 74.5% of respondents, have employees in more than 10 countries. Based on the qualitative responses, it appears that many large companies have employees in at least 30 countries.

Global Locations. Over 70% of respondents have at least hundreds of locations and nearly 30% have thousands of locations.

Divisions and Subsidiaries. Over half of respondents have five divisions or less, and nearly 30% have more than 11 divisions. Exactly 50% of respondents have more than 11 subsidiaries.

Determining Cash Compensation of All Employees

The survey asked whether total annual cash compensation of all employees globally on an individualized basis was easily obtainable by the U.S. compensation staff, and 84% of respondents indicated that the information was not easily obtainable. Of that 84%, over 70% indicated that gathering that information would be very difficult. An illustration of some of the comments issuers made in explaining why the determination would be difficult include:

• “Cash comp[ensation] in the US and Canada can come from our [human resources information system]. For the other 30 countries, we would have to go to each local payroll and define the types of pay we would need for each employee (which I’m sure are all coded differently in each different payroll system). And that would only give us base & incentives and some other special payments.”

• We “currently have approximately 3 dozen payroll systems/vendors globally. Not all locations have a centralized HR shop either, so we would have to devote a lot of people/time.”

• “90 different payrolls . . . in different systems or statutory [requirements]; currency conversions difficult. Very difficult as cash compensation has different components in different markets.”


• “Cash compensation is handled by each country individually with little oversight in terms of delivery between the local HR staff and the country-specific payroll system. To get accurate data, we would likely have to work with every payroll vendor globally to request records for the prior year from which we can generate the total cash compensation figures for employees. Since payroll systems are outsourced outside of North America, this would likely be both time consuming and costly to complete. Depending on the definition of cash compensation, it may be next to impossible to certify that the information is accurate across all the countries in which we pay employees.”

• “Small populations spread across the world with varying international pay plans.”

• “The level of economic development varies significantly amongst our sites (developing to modern), with significant differences in amounts and forms of compensation, as well as currency values, economic/tax/social structures.”

• “First, we have no existing way of calculating the annual total compensation of every employee around the world. … We have 101 payroll systems worldwide. We have approximately 3,600 international assignees, and these assignees get paid in two places (home and host country). … We have six countries that use non-calendar tax years. Not only do we have numerous part-time employees, but we have many different employee types; e.g., multiple types of supplemental employees, different types of inactive employees, and employees who work for wholly owned and less than wholly owned subsidiaries. In addition to the challenge of making an accurate calculation, privacy regulations in certain areas around the world make the data difficult to even obtain. Also, it would be impossible to anonymize data for international assignees because they end up with an identification number for both their home and host country, so there isn’t a way to tie the employee to both payments.”

Even companies that are integrating their information systems globally report that they would continue to have to do manual calculations:

• “We operate on 30+ separate payroll systems today. We are implementing a global [human resources information system] in 2013, which will give us global access to salary information. We will still have separate payroll systems, so reporting actual earnings for all cash compensation (including earned bonus and sales incentives) would continue to require manual consolidation. Consolidating the global database of pay for all employees would take support from 30 country payroll teams.”


Annual Cash Compensation Would Require Manual Calculations for Most Firms

Further reinforcing the notion that developing the information to calculate the median compensation using sampling would be difficult is that 76.1% of respondents indicated that they would have to have a team manually calculate at least some of the data and 47.8% indicated that calculation would be predominantly manual based upon the sampling requirements. Many individual respondents stated something similar to “some of these data are in electronic payroll systems while others would need to be manually pulled together.” Another respondent stated:

“We have a majority of employees in a centralized data base, but not all incentive plans are included so this would be manual. We would collect manual data for salary compensation from several countries. In addition, without a clear definition from the SEC by country of what's included in cash compensation, collecting that information would also be difficult.”

Half of Respondents Would Need at Least Three Months to Calculate Median Employee Pay

Nearly half of all respondents – 49.5% -- stated that it would take their companies at least three months to calculate median employee compensation, and nearly 20% – 18.7% - indicated that it would take their companies five months or more. The responses raise a doubt that many issuers would be able to calculate this information on an annual basis in time to insert it in the proxy statement, let alone multiple times per year in various SEC filings.

One subscriber explained that the availability of year-end data is one impediment:

• “Note that the raw compensation data will generally not be available for U.S. employees before January 31. For companies with international employees who are not subject to U.S. income taxes, it is unclear if, and to what extent, foreign payroll providers will have the necessary information or agree to make it available within the time required for a U.S. company to do the pay ratio calculations. For companies that have an annual shareholder meeting during the March-May “proxy season,” it is likely to be difficult and expensive to accumulate and evaluate the needed data in time to mail proxy materials, even if they do not have international operations.”

As with responses to the earlier questions, a common explanation for why the collection would take so long was that “the manual collection of the data would be the most time consuming.”

Issuers Believe That Sampling Would Not Save Time and Would Add Complexity

As part of the Center’s survey, many issuers expressed strong opinions against the use of statistical sampling to implement the pay ratio disclosure because they do not believe it would save time while adding to the complexity of making the calculation.

One respondent noted that “based on the size and complexity of our global business, we are skeptical that the use of statistical sampling will lessen in any meaningful way the administrative


burden of this exercise.” One issuer, in an industry where the vast majority of the workforce consists of hourly employees, queried how a meaningful statistical sample can be developed that identifies the median employee if cash compensation is the only measure used. Another issuer noted that “very few employers would have the in-house expertise needed to conduct a sophisticated sampling model.” This issuer stated,“it could easily be just as expensive and time consuming to perform the proposed sampling . . . as it would be to just perform all of the calculations for all employees (or a large subset).”

Another issuer explained the problems with sampling in this way: “[t]o a Compensation Professional, the statistical sampling approach does not make sense. From a Comp[ensation] perspective, you just wouldn't find the ’median person’ and then determine total pay. I hope this means that the data will be ‘furnished’ not ‘filed’ - because it will just be an estimate.”

In the end, sampling creates as many issues as it solves. If the SEC were to use an estimate to reach the median compensation, it should consider allowing companies to compare CEO pay to a median industry compensation number, such as is published by the Bureau of Labor Statistics, as is discussed at greater length below.

Issuers Question Whether Total Cash Compensation Is an Appropriate Proxy for Total Compensation of Employees Outside the U.S., Given Significant Global Differences

Based on the feedback from our Subscribers, the Center believes that using only cash compensation for determining the median compensation of all employees globally, as suggested by the AFL-CIO, could result in an inaccurate determination. A chief reason for this is because “in kind” contributions make up a substantial part of compensation in certain parts of the world due to tax considerations, custom and culture and are considered a substitute for cash. For example, one Subscriber noted that the SEC would “need a strict definition of cash compensation, especially in countries where non-cash remuneration is a significant part and substitute for cash compensation.” This is different than in the U.S., where cash compensation is the predominant form of pay for many employees.

Another Subscriber explained that “total compensation packages vary across the globe in both the elements included and the manner of delivery (cash, reimbursement, salary sacrifice [in which employees trade cash for certain benefits or goods], novated car leases [leases paid for by the employer that transfer ownership of the vehicle to the employee], etc.) Determining a consistent method to perform the calculations per the SEC proxy guidance on international total compensation packages would be complex. A further concern may be the degree to which individual country tax/pay regulations may affect the ability to perform U.S.-based calculations.”

Just as using total cash compensation would skew the median calculation for U.S. employees by eliminating forms of compensation that more highly paid employees receive (such as stock options), it is even more problematic abroad where other benefits are often substituted for cash.


The Commission Should Be Willing to Consider Other, More Flexible Approaches to Developing a Pay Ratio

The Commission is considering using an estimate to determine the median compensation of all employees, consistent with the way the U.S. Census Bureau and the Department of Labor’s Bureau of Labor Statistics use statistical sampling to estimate median income of various populations.3 However, if the SEC is willing to use an estimate to approximate median employee compensation through statistical sampling, why require companies to divert valuable time and resources to collecting an estimate of median employee pay when similar estimates are currently available? An approximate pay ratio can currently be determined by comparing the total compensation of the CEO, which is already publicly available,4 to employee earnings that are also currently publicly available through the Department of Labor’s Bureau of Labor Statistics (“BLS”). In fact, companies could customize the ratio to their primary industries or create a blended employee pay rate based on existing BLS data that comparatively would be far more cost effective than the global data gathering exercise explained above and would support efficiency and business growth.

Given the significant concerns that the pay ratio provision raises, and SEC consideration of alternatives, the Center urges the SEC to take the necessary time to develop more a more flexible approach to the pay ratio issue. In addition to taking the time to properly craft these regulations, the SEC should consider the need to provide companies with sufficient time to implement the pay ratio rule once it is finalized. As discussed, aggregating payroll systems will be a significant undertaking and quite a costly venture and companies should be afforded the time to carefully implement these rules.

3 See, e.g., the definitions that the Department of Labor’s Bureau of Labor Statistics uses to conduct its Occupational Employment Statistics survey, http://www.bls.gov/oes/oes_emp.htm#overview (The Occupational Employment Statistics (OES) program conducts a semi-annual mail survey designed to produce estimates of employment and wages for specific occupations.”).

4 CEO compensation disclosures may be found in a company’s annual proxy statement. It is important to note that the pay ratio mandate does not alter any current disclosures. It simply adds the ratio to all SEC filings.

http://www.bls.gov/oes/oes_emp.htm#overview


Conclusion

The Center appreciates this opportunity to provide additional comments on the implementation and rulemaking related to Section 953(b) of the Dodd-Frank Wall Street Reform and Consumer Protection Act. If you have any questions about the Center’s comments, please do not hesitate to contact me at [email protected].

Sincerely,

Timothy J. Bartl Senior Vice President and General Counsel

cc: Securities and Exchange Commission: Hon. Mary L. Schapiro, Chairman

Hon. Elisse B. Walter, Commissioner Hon. Luis A. Aguilar, Commissioner Hon. Troy A. Paredes, Commissioner

Hon. Daniel Gallagher, Commissioner

Securities and Exchange Commission – Division of Corporation Finance: Ms. Meredith Cross Mr. Lona Nallengara Ms. Paula Dubberly Ms. Felicia Kung

United States Senate Hon. Tim Johnson Hon. Richard Shelby

United States House of Representatives Hon. Spencer Bachus Hon. Barney Frank Hon. Scott Garrett Hon. Nan Hayworth

mailto:[email protected]

MEMORANDUM C11-97 NOVEMBER 11, 2011

TO: Center On Executive Compensation Subscribers

FR: Timothy J. Bartl, Senior Vice President and General Counsel Kendra Kosko, Policy Counsel

RE: Results of Center Quick Survey on Use of Statistical Sampling to Produce the Dodd-Frank Pay Ratio

In response to comments filed by the AFL-CIO suggesting that the SEC should implement the Dodd-Frank pay ratio mandate by requiring companies to determine the median compensation of all employees globally through statistical sampling, the Center On Executive Compensation conducted a quick survey of its Subscribers as well as Members of HR Policy Association from October 31, 2011 to November 7, 2011. Overall, 95 companies responded to the survey. The survey was intended to demonstrate the difficulty companies would have with gathering the data required to calculate the pay ratio, even through sampling, which was apparently intended to minimize compliance burdens.

Overall, the survey found that identifying the median employee through the use of sampling would not make the pay ratio calculation much easier. As one respondent noted, “based on the size and complexity of our global business, we are skeptical that the use of statistical sampling will lessen in any meaningful way the administrative burden of this exercise.” We learned that statistical sampling may, in fact, exacerbate the problems involved in calculating the pay ratio since most companies would not be able to handle statistical sampling in-house.

Several companies provided anecdotal responses that help put the overall results into context because the questions were narrowly tailored for this survey. Accordingly, we encourage you to review the comprehensive survey results, which include specific company responses (that have been edited to preserve anonymity of the company). A brief summary is outlined below and followed by the comprehensive survey results.

Overview of Results

• Over three quarters of respondents (78.7%) have over 10,000 employees globally and roughly the same percentage of respondents (74.5%) is located in more than 10 countries.

• The majority of respondents (84%) indicated that it would not be easy to obtain the information necessary to calculate the pay ratio.

• More than three quarters of respondents (76.1%) responded that they would have to manually calculate some, if not all, of the information to determine the pay ratio. Nearly 50 percent said that manual calculation would be the predominant approach.

• While many respondents noted they were guessing, almost half of respondents (49.5%) said it would take at least three months or more to calculate the pay ratio.

Thanks again to all companies who took the time to respond to the survey. We continue to lobby for the repeal of the Dodd-Frank pay ratio provision and will provide updates, as necessary. Please don’t hesitate to contact us should you have any questions regarding this survey.

mblanchard

New Header

Quick Survey on Quick Survey on the Use of Statistical Sampling to Produce the Dodd-Frank Pay Ratio

Survey on Use of Statistical Sampling to Produce the Dodd-Frank Pay Ratio

1. How many employees do you employ globally? Response Percentage

Response Count

A. 10,000 or less 21.3% 20 B. 10,001 to 50,000 41.5% 39 C. 50,001 to 100,000 13.8% 13 D. More than 100,000 23.4% 22

answered question 94 skipped question 1

2. In how many countries are you located? Response Percentage

Response Count

a. 1 6.4% 6 b. 2-5 9.6% 9 c. 6-10 9.6% 9 d. More than 10 74.5% 70


3. If possible, how many locations do you have worldwide? Response Percentage

Response Count

a. 100 or less 27.9% 24 b. Hundreds 43.0% 37 c. Thousands 29.1% 25


4. How many divisions does your company have? Response Percentage

Response Count

a. 5 or less 54.1% 46 b. 6-10 16.5% 14 c. More than 10 29.4% 25


5. How many subsidiaries does your company have? Response Percentage

Response Count

a. 0 17.2% 11 b. 1-5 21.9% 14 c. 6-10 10.9% 7 d. More than 10 50.0% 32


Page 1


6. Is the total annual cash compensation of all employees worldwide currently easily obtainable by the U.S. compensation staff?

Response Percentage

Response Count

a. Yes 16.0% 15 b. No 84.0% 79


Company Response

Company 14 A. We keep a database from our payroll system of all cash compensation for all employees and convert it on a daily basis.

Company 51 A. Only two countries. It would take some time, but if just cash compensation, we could get to that in a few days.

Company 87 B. All compensation is available but conversions would need to be made for currencies in many countries and other adjustment would need to be made.

Company 95 B. The sum total is obtainable, but getting each individual’s total annual cash comp by person would be very time consuming and difficult.

7. If your answer to number 6 was no, how difficult would it be to develop the cash compensation of all employees worldwide?

Response Percentage

Response Count

a. Not difficult (please explain) 1.3% 1 b. Somewhat difficult (please explain) 27.8% 22 c. Very difficult (please explain) 70.9% 56


Company Response Company 1 C. Every location has its own payroll system.

Company 2 C. Multiple payrolls, and central database is not kept up to date and is often fed by spreadsheets. Only covers about 90% of employees.

Company 3 C. Countless hours and resources necessary.

Company 5

C. Cash compensation in the US and Canada can come from our human resources information system. For the other 30 countries, we would have to go to each local payroll and define the types of pay we would need for each employee (which I’m sure are all coded differently in each different payroll system). And that would only give us base & incentives and some other special payments.

Company 6 C. This information must be obtained from disparate payroll systems around the world.

Company 7 C. We are working toward a global human resources information system platform, but currently have approximately 3 dozen payroll systems/vendors globally. Not all locations have a centralized HR shop either, so we would have to devote a lot of people/time.

Company 8 C. Manual exercise. Company 9 C. We have upwards of 50 independent payroll systems and none of them communicate. Company 10 C. We have no global system. Many roll-ups [total calculations] are done by spreadsheets in many countries.

Company 11 B. Total cash compensation figures are not housed within our HRIS system. Accordingly, to obtain the complete cash compensation picture we’d need to either (1) obtain and aggregate sales incentive compensation data from our sales organizations or (2) query our payroll system.

Company 12 C. 90 different payrolls and pension values in different systems or statutory; currency conversions difficult. Very difficult as cash compensation has different components in different markets.

Company 13 C. We would have to extract the data from local payroll systems in each country and then apply whatever rules the SEC comes up with. We expect this would be very time consuming and fraught with the possibility of error.

Company 15 C. Compensation data for non-U.S. employees will be difficult to gather due to the lack of a centralized global payroll system.

Company 16 B. Would have to gather information from a central repository plus from local payrolls in each country.

Company 17

B. Although we currently have access to global employee total cash compensation, there is manual intervention required for recently acquired companies that have not yet implemented the Global human resources information system in addition to country/government rules and regulations interpretation based on the definition of total cash compensation.

Page 2


Company 18 B. We’d have to coordinate with all the local payrolls for each location and aggregate the list converting to US dollars to determine the cash compensation for all employees.

Company 19 C. We employ approximately 180,000 employees, the majority of whom work flexible, part-time schedules. In addition, many of our employees earn tips in addition to their base wages.

Company 20 C. We have multiple subsidiaries and divisions in multiple companies and no central location to accumulate the data.

Company 21

C. As we do not have a centralized payroll system we would need to rely on the data being provided by our local entities, which in some cases may have more than one payroll system themselves. Further, a somewhat significant portion of our population is part-time, which would require additional analysis to determine total cash compensation.

Company 23 C. We primarily have only domestic employees in our human resources information system software. We would have to reach out to all of our locations and compile the cash compensation data, pick a standardized date, etc.

Company 24

C. Cash compensation is handled by each country individually with little oversight in terms of delivery between the local HR staff and the country-specific payroll system. To get accurate data, we would likely have to work with every payroll vendor globally to request records for the prior year from which we can generate the total cash compensation figures for employees. Since payroll systems are outsourced outside of North America, this would likely be both time consuming and costly to complete. Depending on the definition of cash compensation, it may be next to impossible to certify that the information is accurate across all the countries in which we pay employees.

Company 25

B. Our industry has a high rate of part-time employees and a high rate of turnover so we would need to determine the handling of those populations. After culling the data to the appropriate population, we would need to identify all of the elements of “compensation” in each country, make a determination of how it relates to executive compensation. It would likely require the installation of an accounting or consulting firm to ensure that our understanding of the requirement and the treatment of all elements of compensation is accurate. The process would take months to accomplish, at great expense, and in the end I’m not sure what level of comfort I would have that we have selected the right median employee.

Company 26 C. This will be difficult to obtain because we have multiple payroll systems and pay (cash compensation) is defined and structured differently for some countries.

Company 29 C. No global HRIS. We would have to go location by location and country by country.

Company 30 B. We do not have a global compensation program. All work would be manual.

Company 31 C. Small populations spread across the world with varying international pay plans. Company 32 C. No central payroll system; 13+ systems for collecting data. Company 34 C. We do not have a common HRIS system. It would be thousands of man hours using spreadsheets.

Company 35 C. Very difficult right now because we do not have a centralized HR system and not all cash bonus information is housed in our systems today.

Company 36 C. We would have to analyze the output of all of the different payroll systems worldwide to determine the cash compensation of the employees paid in from that system.

Company 37

C. We use Peoplesoft worldwide which would be easier to get base wages from; however, a few countries are not utilizing the system yet and tracking other forms of compensation like housing, meals, etc. are not kept in any global available HRIS system; further, I am not clear how entry level wages in SriLanka compared to the US CEO helps anyone make a decision on anything.

Company 38 C. We have 114 different ERP systems and at least 200 payroll systems in operation globally.

Company 40 C. This would require collecting data using spreadsheets which would require corresponding with local HR and Finance employees in each country and subsidiary.

Company 41 C. The level of economic development varies significantly amongst our sites (developing to modern), with significant differences in amounts and forms of compensation, as well as currency values, economic/tax/social structures.

Company 42 B. Cash compensation could be pulled out of SAP with a little effort. Benefits would be manual.

Company 43 C. We have different HR systems/platforms used in each region of the world. Countries may use different approaches to compensation based on legal and local issues. We’d have to figure out a way to make this an apples to apples comparison.

Company 44 C. Although we can gather notional data from our international sites, we do not currently have consistent global use of our HRIS system and many local payroll providers are utilized.

Company 45 C. No centralized payroll system, non-US definitions of pay vary by country.

Page 3


Company 46 C. No centralized payroll system so information needs to be obtained from each location. In addition, need a strict definition of cash compensation, especially in countries where non-cash remuneration is a significant part and substitute for cash compensation.

Company 48 C. Many countries are on different systems that are not integrated.

Company 50 C. We do not have a global HRIS, so all data would need to be collected manually, assisted by a great software tool by Microsoft called Excel.

Company 52 C. We have many part-time employees, and a very fluid workforce with significant turnover.

Company 53 B. Staff in acquired companies not available in our global HRIS system. Would require a manual effort, but for a couple companies.

Company 54 C. Given the number of countries in which we operate, disparate payroll systems and unique incentive programs (e.g. many countries have their own sales incentive programs), the collection of the data needed to produce this calculation will continue to be an extremely labor and time intensive process.

Company 56

B. We operate on 30+ separate payroll systems today. We are implementing a global human resources information system in 2013, which will give us global access to salary information. We will still have separate payroll systems, so reporting actual earnings for all cash compensation (including earned bonus and sales incentives) would continue to require manual consolidation. Consolidating the global database of pay for all employees would take support from 30 t ll t Company 57 B. We don't have a central payroll system. Would have to collect from each entity.

Company 58 C. We would need to require information from field offices. Each country has a different definition of cash compensation. A high percent of international based employees are time/labor and/or sales employees where there is no supporting system to report compensation.

Company 61 C. Countries operate independently. Subsidiaries operate their own payroll systems (in-house or outsourced) around the world. Data privacy laws make it difficult to access personal information on employees outside the U.S.

Company 62 C. There is no standard payroll or compensation system and local payrolls, benefits, practices and codes are in local languages.

Company 63 C. While we can ascertain base salary, all other compensation would be very difficult to gather.

Company 64

C. We have no centralized reporting mechanism for compensation at the individual employee level globally. Capturing annual salary by employee at any given time would be a manual process compounded by a number of factors including the following: • Employee Turnover • Employee changes in pay – transfers, promotions, full-time to part-time etc. • Definition of annual salary – (e.g. inclusion of overtime, holiday, sick leave pay, salary reduction etc.) • Currency conversion – fluctuation of exchange rates

Company 66 C. Each country maintains its own records on whatever system they have. The systems are all different. We are decentralized. And, of course, multiple currencies.

Company 68 C. Not all countries/regions on same HR management systems, HR is managed regionally, so do not have an HR representative in every country where we may have an employee.

Company 73 B. Based on the AFL-CIO proposal, we have several areas that would have different annualization rules based on the number of hours worked (e.g. 37.5, 40, 42, 45 hours per week). Would have to be manually calculated outside of current human resources information system.

Company 76 B. Quarterly bonuses in our company are difficult to extract/summarize. Company 80 B. Would need to collect some data manually and adjust for currency.

Company 81 C. Multiple payrolls, many of which are run by third party vendors and how is “Cash Compensation” defined in all of the countries outside the US, as there are many elements to pay in most countries outisde the US.

Company 82 C. We do not have a centralized payroll system nor access to each of the individual payroll systems. We are particularly concerned about capturing the wage compensation since this data has not been collected globally.

Company 84 C. Annual salary is readily available in PeopleSoft but would not include additional earnings such as bonus, overtime, shift pay, etc. We would need to reach out to 57 different countries (some with more than one payroll system) and obtain that data from each country and consolidate it.

Company 85 A. We would have to reach out and ask for it.

Company 86 B. Work-intensive as it would involve soliciting information from basically stand alone businesses, estimate short-and long-term incentives, and currency converting for reporting purposes.

Page 4


Company 88 With multiple locations, multiple payrolls, multiple currencies, full-time, part-time - it will be difficult.

Company 89

C. First, we have no existing way of calculating the annual total compensation of every employee around the world. This process for five individuals is already extremely time consuming and labor intensive. We have 101 payroll systems worldwide. We have approximately 3,600 international assignees, and these assignees get paid in two places (home and host country). We have 74 different pension plans worldwide. Also, what is reported as taxable compensation and the level of retirement benefits are very different around the world. We have six countries that use non-calendar tax years. Not only do we have numerous part-time employees, but we have many different employees types; e.g., multiple types of supplemental employees, different types of inactive employees, and employees who work for wholly owned and less than wholly owned subsidiaries. In addition to the challenge of making an accurate calculation, privacy regulations in certain areas around the world make the data difficult to even obtain. Also, it would be impossible to anonymize data for international assignees because they end up with an identification number for both their home and host country, so there isn’t a way to tie the employee to both payments.

Company 91 C. We currently do not have a global human resources information system, nor common short-term Incentive programs that have been globally identified or tracked.

Company 92 B. Somewhat. We do not have HR systems in every country. Plus there are some countries in Europe that we are not allowed to transmit data involving personally identifiable information outside the continent.

Company 94 B. Data would have to be gathered from each country and manually integrated.

Company 95 B. We would need to pull data from a number of data bases around the world and audit for accuracy. This would take many man hours of time.

8. How would you calculate the cash compensation of your employees (or a sample of employees) worldwide?

Response Percentage

Response Count

a. We have a centralized payroll computer system in place that we would use. 10.9% 10 b. We have payroll systems in each location or regionally that we would use to aggregate the data.

28.3% 26

c. We would have to have a team to manually collect the data based on the sampling requirements.

47.8% 44

d. Other. 13.0% 12 answered question 92

skipped question 3

Company Response Company 5 B. But these are all independent so it would be extremely difficult to get this data. Company 15 D. Both B and C.

Company 17 A. We have a centralized global system in place that we would use for those countries/organizations who have fully implemented. However, individual country/organization contact would be required to confirm accuracy of data based on local situation and required definition of total cash compensastion analysis.

Company 21 B. While we have payroll systems in each location to provide us the data, we would still need to perform a manual process to collect and aggregate the data.

Company 23 C. We’d probably have to compile all data in spreadsheets from the locations and have a team analyze it.

Company 25 B. We do have payroll systems, or third-party payroll providers, regionally. What they might charge us to get the data in the format we need could be another added expense.

Company 27 D. Both A and B. Company 28 D. Likely a combination of b and c. We do not yet have a global human resources information system or

Company 31 C. Manual compilation, and with EU data privacy and other privacy limitations, collection will be made even more difficult.

Company 35 D. We would likely have a centralized compensation team collect from the regions and consolidate.

Company 36 B. As above, we have systems, but each one includes country-specific data elements. We would have to define cash compensation for each payroll system to know what we were getting.

Company 37 C. We absolutely will need to do this manually.

Company 45 C. Some of this data is in electronic payroll systems, while some would need to be manually pulled together.

Company 47 D. We would rely on our human resources information system, and where gaps, go to local payroll systems.

Company 49 D. We have a centralized human resources information system that captures salary and bonus information. It's not centralized payroll specifically, but it is a centralized HRIS that captures this data, so it would provide a very reasonable/solid estimate.

Page 5


Company 58 D. In some locations, we have payroll systems, but the systems are from different vendors. However, there are others that we would need to manually collect data.

Company 62 D. It would be a combination of b and c to get SEC defined total compensation.

Company 65

C. We have a majority of employees in a centralized database, but not all incentive plans are included so this would be manual. We would collect manual data for salary compensation from several countries. In addition, without a clear definition from the SEC by country of what's included in cash compensation, collecting that information would also be difficult.

Company 66 B. We would need an HR or payroll representative in each country to pull the data off their system and send it to us at Headquarters to aggregate.

Company 71 B. We have a payroll system for the U.S. and a payroll system for Canada.

Company 74

C. The employees in the U.S., the employees in Canada, and the employees in China are each paid through a separate third-party payroll system provider. We would have to get the appropriate information from each provider, and then have that data sorted and aggregated. Since the three payroll systems are not compatible, it is unclear whether we could develop a computer program to handle this problem, or whether we would have to do the entire calculation through spreadsheets and similar, more manual processes.

Company 77 C. Not sure that we would get funding to assign a team, however, so this would have to be handled by our compensation professionals in addition to regular activities.

Company 80 B. Some of our payroll systems are based on spreadsheets in smaller geographies. We would need to cross check against our HR system (SAP) which does not quite have full global data.

Company 81 C. We may be able to pull together data from a centralized payroll system in certain countries/regions, but we would ultimately have to pull the data together and analyze it manually.

Company 86 C. We would have to have a team to manually collect the data based on the sampling requirements as some locations are remote and others will require interpretation as to what constitutes "cash compensation."

Company 89 D. Both b & c, we would need to invest in a special database to handle this new process. Company 92 C. Outside of North and South America it would be manual.

Page 6


9. Given the summary of the proposal for the use of statistical sampling of cash compensation for calculating the median pay of employees globally, how long do you estimate it would take your company to identify the median employee (please explain)?

Response Percentage

Response Count

a. Less than one month 18.7% 17 b. 1-2 months 31.9% 29 c. 3-4 months 30.8% 28 d. 5-6 months 8.8% 8 e. More than 6 months 9.9% 9


Company Response Company 1 C. We would have to bother payroll/HR administrators at every location to obtain the data. Company 2 C. Collecting key data on all employees. Company 4 A. Obtaining data would take 1 week. Analyzing data would take 1 week.

Company 7

C. Form a team, get the team up-to-speed on the task, collect data (long process), clean data (e.g. make sure we’re looking at just FTE’s if that is the rule), confirm data with local management and local payroll, finalize data, analyze, submit. And that is only if there is a team devoted solely to the task – as competing day-to-day priorities would have to be considered. Might even take longer.

Company 15 B. Difficult to estimate as it depends on the number of people on the project. Company 17 C. Estimate 3 months based on necessary reporting and individual country/organization interface required. Company 20 C. It is the manual collecting of the data that would be the most time consuming.

Company 21 C. This estimate is based on the amount of time it would take for each of our locations to determine the total cash compensation for our all employees, including part-time, and then aggregating the data from Corporate.

Company 23 B. It would probably take us about 1-2 months due to the level of detail required and the fact that we don’t have the resources to take on this project in addition to other work that we’re already performing.

Company 26 B. The timing considers working with the other payroll locations and manually collecting data. Company 29 B. If HR and Finance could free up enough local resources to focus on this. Company 31 D. 5-6 months for a start-up, then 3-4 months on an ongoing basis. Company 34 E. Definitely more than 6 months.

Company 35 B. The timing really depends on the time of year that the calculations are done and the availability of resources and data.

Company 36 B. Simply a guess.

Company 37 E. Likely six months and then I would question the accuracy; the question of time versus value remains even if we were 100% accurate.

Company 40 E. The sheer volume of interactions will take time. Not every country structures compensation the same way as the U.S. does. We will have stragglers and folks who fill out spreadsheets incorrectly. So, there will be a lot of scrap and rework.

Company 44

E. Selected due to our decentralized systems and the significant degree of effort required to coordinate the collection of data and the subsequent analysis of the information. Additionally, total compensation packages vary across the globe in both the elements included and the manner of delivery (cash, reimbursement, salary sacrifice, novated car leases, etc.) Determining a consistent method to perform the calculations per the SEC proxy guidance on international total compensation packages would be complex. A further concern may be the degree to which individual country tax/pay regulations may affect the ability to perform U.S.-based calculations.

Company 45 C. The approach to statistically sampling is untried and unspecified; in addition, sampling appears to still require data from all global locations.

Company 50 D. Depends on how many resources we put on it.

Company 52 B. Almost impossible to be accurate and our median employee would be relatively low paid since the majority of our employees are hourly store employees.

Company 55 A. Simple definition of pay.

Company 56 B. Elapsed time likely to be 1-2 months to define global requirement, extract data, consolidate & perform analysis.

Company 61

C. The calculation of pay to determine the median is probably less of an issue than defining and identifying the sample population and the methodology for calculating pay. Annual cash compensation would have to be further defined - is it all the cash that is paid to employee in a given year, or is it the 12 times monthly base pay plus annual short-term incentives, or some other definition?

Page 7


Company 64 B. Minimum 1-2 months, assuming dedicated, resource-intensive effort. Company 65 C. The entire process, including lead time to the different locations would result in a 3-4 month period. Company 66 A. It can probably be done in less than a month if it is a top priority and we can allocate the resources.

Company 71 A. The compensation team can compile the data in the U.S. payroll system and have Canada Payroll run the same data for the Canadian coworkers. This data can be put together and analyzed within a week.

Company 72 B. At least 1-2 months with external consulting assistance. Data would need to be manually collected from multiple different systems, working through various affiliate personnel, country specific definitions of pay (base pay, fixed pay, year end gratuity, retention payments, company contributions/match, etc.).

Company 74 B. Given the lack of detail at this point, it is hard to say, but I would expect it would take one to two months to determine the median employee, even if we were looking at only cash compensation.

Company 75 A. I think we could do this fairly quickly.

Company 80 B. Even with a sample it is a lot of work to calculate the total compensation per the SEC's rules and it still doesn't make sense depending upon the level of the "median" employee.

Company 81 B. If one person focused on this exclusively, we might be able to get it done in less than 1 month, but the logistics alone of gathering data in all of the world areas would be horrible.

Company 82 D. We anticipate it will take a significant amount of time just to understand what data to collect and where to get it from. After the first year, we anticipate less than 5-6 months.

Company 86 B. 1-2 months if not longer depending on the response time of some of our remote locations.

Company 93 A. Less than a month if using legacy HR systems. 3-4 months if using payroll systems to capture actual cash compensation paid.

Page 8


10. General Comments

Company 7

For a US-only company, determining a “median” employee might be somewhat feasible. But for a global company, it seems that a “median employee” figure has the potential to be a very convoluted number which could make any analysis less meaningful. We are building a global base pay structure, but with pay ranges based on local market data. So, trying to compare a “median employee” in a European country with a unique pay structure to a “median employee” in an Asian country with a completely different structure to a “median employee” in the US, and then determining one global “median employee," it starts to fall apart.

Company 20 The issue still is one of questioning the usefulness of the underlying idea. Therefore, sampling as a way to help achieve the outcome helps potentially the administration side but seems to facilitate a bad idea (e.g., time spent producing useless information).

Company 35 We are still in the camp of asking what value this provides to shareholders vs. the cost to the company.

Company 36

What makes anyone think that the median employee on cash compensation is the median employee on total compensation when looking across the globe? Given the different compensation and benefits structures, markets, regulations, taxes, etc. from country to country, it is easy to imagine that the value calculated this way could be off + or - 20-30% or even more. If somebody wants to claim that this doesn't make a difference, I would ask if it is OK to apply the same standard of accuracy to the CEO total compensation? I am guessing not.

Company 51 We aren't too bad because the bulk of our employees are in this country. If they wanted to calculate beyond just base pay, that would be quite a bit more difficult--adding equity, benefits, pension, etc. into the equation.

Company 54 Based on the size and complexity of our global business, we are skeptical that the use of statistical sampling will lessen in any meaningful way the administrative burden of this "exercise."

Company 62

1. In some countries people are on short term work contracts of 3 – 6 months – how would they be handled? 2. In some countries a significant number of workers are contracted through agencies for fairly long periods of time – how would they be treated? 3. Fundamentally, by using the global workforce the results across companies are not comparable. The percentage for a CEO with a large workforce in India would be different than one with a large workforce in Germany even if both CEO’s had the exact same compensation and same number of US employees.

Company 73

Another consideration that may be inherent to us and potentially other retail organizations is that a very large percentage of retail associate populations are store level associates (almost 90%) who are neither bonus or grant award eligible. Just taking a statistical sample of all associates does not address the importance of who the “median employee” is as a comparison to CEO total compensation or make the calculation easier.

Company 74

Note that the raw compensation data will generally not be available for U.S. employees before January 31. For companies with international employees who are not subject to U.S. income taxes, it is unclear if, and to what extent, foreign payroll providers will have the necessary information or agree to make it available within the time required for a U.S. company to do the pay ratio calculations. For companies that have an annual shareholder meeting during the March-May “proxy season”, it is likely to be difficult and expensive to accumulate and evaluate the needed data in time to mail proxy materials, even if they do not have international operations. How is “cash compensation” defined? Some cash payments to foreign employees may or may not be treated as compensation for U.S. purposes or as defined in the proxy disclosure rules, and foreign payroll providers may be unable or unwilling to adjust the data that they provide to accommodate whatever rule the SEC decides upon. Similarly, if the median employee is a foreign employee, companies may have difficulty in “translating” the employee’s compensation to match the SEC proxy disclosure rules. Very few employers would have the in-house expertise needed to conduct a sophisticated sampling model. Said differently, it could easily be just as expensive and time consuming to perform the proposed sampling (followed by determining that employee’s total compensation under the proxy rules) as it would be to just perform all of the calculations for all employees (or a large subset that is designed to pick up the “median” employee). What good would sampling do, besides complicate matters? If a company can manage to get the information for “cash compensation” for all of its U.S. and international full-time and part-time workers who were employed at any time during the year in question, it should be relatively easy to simply arrange them in “total cash compensation” order to pick the median employee.

Page 9


To a compensation professional, the statistical sampling approach does not make sense. From a compensation perspective, you just wouldn't find the "median person" and then determine total pay. I hope this means that the

Company 77 data will be "furnished" not "filed" - because it will just be an estimate. Limit the information to the US only and only on TDC. It is the addition of international data and the other elements from the SCT that makes this an impossible feat (and useless data).

Page 10

ms. elizabeth m. murphy secretary washington, dc …...1 november 11, 2011 ms. elizabeth m. murphy...

Documents